© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Joseph Bloom · Feature Splitting for GPT2-Small
    5. GPT2-Small
    7. Residual Feature Splitting 12288
    9. 8-RES_FS12288-JB
    11. 10510
    Prev
    Next
    INDEX
    Explanations

    instructions or step-by-step procedures

    oai_token-act-pair · gpt-3.5-turboTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs12288-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_12288
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    12,288
    Data Type
    torch.float32
    Hook Point
    blocks.8.hook_resid_pre
    Architecture
    standard
    Context Size
    128
    Dataset
    Skylion007/openwebtext
    Hook Point Layer
    8
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Neuron Alignment
    Index
    Value
    % of L₁
    519
    +0.10
    0.5%
    194
    +0.09
    0.4%
    536
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    519
    +0.10
    0.05
    354
    +0.09
    0.04
    566
    +0.09
    0.04
    Negative Logits
    Therefore
    -0.80
     Therefore
    -0.72
    inventoryQuantity
    -0.69
    Various
    -0.61
    Recently
    -0.60
    govtrack
    -0.58
    TL
    -0.58
     )]
    -0.57
     Due
    -0.57
    malink
    -0.57
    POSITIVE LOGITS
     prest
    1.11
     instantly
    0.94
     vo
    0.85
     magically
    0.84
     reap
    0.83
     explodes
    0.75
    'll
    0.75
     invariably
    0.74
     shalt
    0.70
    arts
    0.66
    Activations Density 0.481%

    No Known Activations